Quickstart

1. Install

curl -LsSf https://getanalog.io/install.sh | sh

Windows, Homebrew, pipx, and pip routes: Install.

2. Sign in

analog signup   # new account: creates it and connects this terminal (invite code required)
analog login    # existing account, or another machine

One or the other, not both — your browser opens to confirm. The credential lands in a local store that the CLI, SDK, and MCP server all read automatically — nothing to copy into a config file.

3. Extract a page

analog get https://quotes.toscrape.com/js/

The result prints as a short preview — sections, record counts, field names, and anything Analog disclosed about the page — plus a handle (like 20260719-k7m2p9) that re-opens the saved result any time, with no re-fetch:

analog open <handle>                # the preview again
analog export <handle> -f csv       # records as csv / json / yaml / md

4. The same call in Python

from analog import analog

result = analog("https://quotes.toscrape.com/js/")
section = result.structured_content[0]
print(section.records[:3])

Next

• Fetching — pagination (--pages N), load-more buttons, per-item expansion.
• Results — query, filter, diff, and export saved results.
• MCP server — use Analog from Claude, Cursor, or any MCP client.
• No account yet? analog assess <url> (the fit check) and analog get <url> --mode local (page markdown, converted entirely on your machine) both work without signing in.